Biomarker pathway heterogeneity of amyloid-positive individuals

Abstract

Introduction: In amyloid-positive individuals, disease-related biomarker heterogeneity is understudied.

Methods: We used Subtype and Stage Inference (SuStaIn) to identify data-driven subtypes among cerebrospinal fluid (CSF) amyloid beta (1-42)-positive individuals from the Alzheimer's Disease Neuroimaging Initiative (ADNIGO/2 [n = 376]). Variables included: CSF phosphorylated tau (p-tau181), hippocampal and whole-brain volume, logical memory (LM), composite Trail Making Test score, and white matter hyperintensity (WMH) volumes. CSF amyloid-negative, apolipoprotein E ε4 non-carrier cognitively unimpaired controls (n = 86) were used to calculate z scores.

Results: One subtype (n = 145) had early LM changes, with later p-tau and WMH changes. A second subtype (n = 88) had early WMH changes, were older, and more hypertensive. A third subtype (n = 100) had early p-tau changes, and reflected typical Alzheimer's disease. Some amyloid positive (n = 43) individuals were similar to the amyloid-negative group.

Discussion: This work identified heterogeneity in individuals who are conventionally considered homogeneous, which is likely driven by co-pathologies including cerebrovascular disease.

Highlights: Data-driven modeling identified marker heterogeneity in amyloid-positive individuals. Heterogeneity reflected Alzheimer's disease-like, vascular-like, and mixed pathology presentations. Some amyloid-positive individuals were more similar to amyloid-negative controls. Vascular pathology plays a key role in understanding heterogeneity in those on the amyloid pathway.

Keywords: Alzheimer's disease; amyloid positive; heterogeneity; mixed dementia; subtype and stage inference; vascular pathology.

FIGURE 1

SuStaIn subtyping depicting biomarker event staging by ordering of standardized scores (one z score [red], two z scores [magenta], three z scores [blue]), with colors between these scales highlighting uncertainty in event staging. WMH used log₂. A dashed line for each subtype is included to represent higher uncertainty in the diagrams, in which number of subjects is ≤ 2 for two consecutive stages. pTau, phosphorylated tau; SuStaIn, subtype and stage inference; WMH, white matter hyperintensities.

FIGURE 2

Biomarkers by SuStaIn stage, across subtype. Individual volumes for subtypes for each stage across subtype are plotted, with a line of best fit. Pairwise correlation coefficient across subtype stage is reported in Table S1 in supporting information. pTau, phosphorylated tau; SuStaIn, subtype and stage inference; WMH, white matter hyperintensities.

FIGURE 3

Pie charts for diagnostic progression by 24 months within baseline diagnostic group (CN or MCI), by subtype group. CN, cognitively normal; MCI, mild cognitive impairment; pTau, phosphorylated tau; WMH, white matter hyperintensities.